voice of the dream

yah, more frequent updates, that's the ticket...
Gonna be a long one today, I think. I really need to work on doing this every night, instead of storing it all up and then exploding...

New design, too, as you might have noticed. 8^)= Let me know about any issues.

just not getting it
I figured we were going to see some crazy ideas as the advertising-funded model broke down, but I didn't see this coming: pay to link schemes. Here's hoping this one dies a quick, painful death.

speaking of crazy ideas
Sounds like the Blogger kids have gotten enough scratch from their IPB (Initial Public Begging) to upgrade a server or two. Wouldn't it be ironic if the most successful funding model on the web turns out to be the donation-driven one that public television and radio have been using for years?

I might be inclined to rant about some of the bitching I've seen about Pyra's request for donations, but Al said it better.

I haven't made a donation yet, cuz I don't use PayPal, but I am going to mail a check tomorrow. I don't use Blogger myself, but I read and enjoy the sites of people that do, so I'm going to contribute.

Oh, and a note to Blogger users -- it might be in your best interest to let the world know what tools you're using, just so people realize how important the continued availability of Blogger is.

also not getting it
It the "continued struggle to control what you do" department, the Register offers up Everything you ever wanted to know about CPRM. If CPRM doesn't mean anything to you, you should definitely have a look at that link.

link graveyard
This probably won't surprise anybody who has tried to get links of weblog archives, but link rot is becoming a serious issue in academic bibliographies. Yet another reason why a central repository for "published" scientific information, like PubMedCentral, is critical for the future of research and academic inquiry.

more porn
Last week I got a thoughtful email from Seth Golub about my Am I Porn or Not? scheme. I finally got around to replying to him today, and about two minutes later, I got an email from him saying that he had posted his email on his site. So, here's my reply (edited somewhat from the form I sent to him:

Seth> It's not that simple, unfortunately.  Humans are able to spot porn,
Seth> but it's not just because of rules we know.  We also have systems for
Seth> extracting complex high level information from images.  We can look at
Seth> a picture and know whether it contains people or people-like things
Seth> and can easily recognize their poses, expressions, and situations.

Yah, I realize that it's not that simple. The idea was (at least
partially) intended to be a bit tongue-in-cheek. Alternatively, you
could argue by the time that enough data has been collected, there
might be some sort of conceptual break-through in computer vision
and/or expert systems design, so that it _will_ be possible to process
high-level rules out of the data.

On the gripping hand, the one thing that I consider the biggest
problem (which nobody has mentioned yet) is that you're relying on
humans to build the initial dataset -- and there are people out there
who would intentionally (for a variety of reasons) input erroneous
data. In fact, since 'porn' is at least partially a social construct,
different people with different backgrounds are going to have
different ideas of what 'porn' constitutes. To do this right, you'll
need to track demographic info about participants, and correlate that
with their choices about what is 'porn' and what isn't. 

Seth> David Forsyth, an associate professor at Berkeley, has published
Seth> several papers on his research into automatically detecting naked
Seth> people in images.

This sort of thing makes me somewhat nervous (not nervous that the
research is being done, but because it's all too easy to see it being
mis-applied). Naked people != porn; sometimes naked people == art, and
sometimes people in porn are completely clothed. As I said above, what
you consider porn is very much a function of your socialization and
your individual sexuality, and I think extracting abstract rules out
of that mess is going to be very, very difficult. 

What I was sort of talking about trying to do would involve the
development of some sort of heuristic; it wouldn't be right all the
time, but it might be better than what we've currently got.

Seth did send me a reply to the above, but I'm not comfortable posting it here without explicit permission; I'm pretty sure it'll appear on aigeek soon.

linux newbies, heads up!
Helped a friend get some stuff set up on his Linux box this weekend. He'd done most of the hard stuff, but was having an issue or two with X and PPP, which I was able to solve. He might find these 10 tips useful, and you might too, if you're just getting started out.

distributed itch scratching
Ian Clarke has a word or two for people bitching about Freenet's ease of use. I think this sort of "piss off or pitch in" message is going to become increasingly more common as use of Open Source software spreads. People need to understand that being given a gift doesn't place the giver under any obligation to the recipient; unfortunately, most people have to be publicly spanked before this sinks in.

dsl deadbeats
Looks like the telco strategy of providing crap service to DSL resellers is starting to yield fruit: two DSL resellers are shutting down. This indirectly hurts Covad, because they're not getting paid for the service they provided, and it directly benefits the telcos, who would like nothing better if all the third-party providers would just go away. Boy, nothing like market forces at work, eh?

right/wrong blindness
A company has purchased a patent on the human genes for red-green color blindness. Aside from my often-stated opposition to gene patenting in general, there's one glaring problem: the title of the press release: "Medical College of Wisconsin Awards ColorMax Patent for Human Color Genes". News flash: the US Patent Office awards patents. What actually happened here was that the Medical College of Wisconsin sold the patent to this company.

your biological future
Red Herring has an interview with Leroy Hood, one of the pioneers of modern biology. He says some interesting stuff, but the interviewer didn't ask any of the tough questions. I would have liked to hear Hood's viewpoints on gene patents and data release. For example, Hood lists four major impacts of the human genome project, but he doesn't mention that all of these impacts were dependent on the open release of data from the project, a practice at odds with the way most biotechs do business.

(Link via the always top-notch snowdeal.org {bio,medical} informatics.)

$bloggers--
Breaching the Web has submerged. Thanks for the links, Kris, and hurry back.

site meta
Far more than you probably wanted to know on the 2000 Genehack stats and search terms wrap-up.